constant current source

[hacknet]

hey,

i was toying around with complimentary feedback pairs and i was wondering if ring of two constant current source was superior to a jfet ccs? i was thinkin of some run off the mill 2n5484s.

thanks a million!

[ilimzn]

In short: YES! Just take care that the ring of two does not oscillate.

[Ouroboros]

Although a simple JFET (or a depletion-mode MOSFET) CCS won't have as high an impedance as the ring-of-two at low frequencies, it will maintain its CCS behavoiur up to a higher frequency. This is because of the 'Miller' capacitance causing NFB that lowers the Z-out of the 'ring' circuit as the frequency increases.

Also the 'ring' circuit hac poor dc performance over temperature, as the current setting is determined by the Vbe of the bottom transistor.

I've seen people using 'ring' constant current loads for long-tailed-pair double-triode phase splitters in valve amps who wonder why the phase splitter performance is so poor at higher frequencies!

[hacknet]

hmm.. are there any specific specification to look for in selecting ring of two transistors?

[capslock]

what configuration do you call a ring CCS?

[hacknet]

#4

[Ouroboros]

Originally posted by capslock
what configuration do you call a ring CCS?

This sort of thing.

[lumanauw]

Anybody knows what constant current source that can work properly only under 0V3? In figures above all needs more than 2 diode/VBE drop to form CCS.

[lineup]

Originally posted by lumanauw
Anybody knows what constant current source that can work properly only under 0V3?
In figures above all needs more than 2 diode/VBE drop to form CCS.

Current Mirrors can have drop lower than 0.2 Volt.

Say you want a low drop current source from negative rail.
To feed a dual NPN differential pair.

You then make a current source from POSITIVE rail
and under NPN pair, at negative rail,
you put a simple (two transistors) current mirror
to bias the differential pair.

To increase precision, if you need very good precise current source,
you might even use small value emitter resistors
in the two transistors in mirror.
Say with only 0.05-0.02 Volt volt drop at chosen current.

This will lower the saturation margin, Vce in mirror transistor (Vce 0.25-0.28, instead of Vce=0.30)
So, if you do not need absolute highest precision,
you should use a current mirror wthout emitter resistors.
To maximize Vce and avoid saturation in mirror transistor.

[lumanauw]

Hi, Lineup,

This is my problem

I wanted to replace R3 (4k7) with CCS to give constant current benchmark for Q3 base.

I don't want to built current mirror between Q1 and Q2's collector, because current mirror there also gives non-constant current source for base of Q3 (Resistor at R3 also non-constant, depending the voltage drop accross R3).

The problem is because Q3 is bipolar. If Q3 is Vmosfet with drop 3V5, I can make transistor CCS based by 2x4148 or led as reference, because the margin is high. But if Q3 is bipolar without any RE, the margin is only 0V6.

Do you have reference, what CCS will fit in position R3? How do IC designers makes CCS inside opamp?